Cognitive Therapy and Research, Vol. 24, No. 4, 2000, pp. 399–418

Think-Aloud, Thought-Listing, and Video-Mediated Recall Procedures in the Assessment of Children’s Self-Talk Jackie Lodge,1 Gail Tripp,1,2 and Diana Kim Harte1

Two studies examined the extent to which different cognitive assessment procedures yield similar data in pre-adolescent children. In Study 1, situationally anxious children reported their self-talk via think-aloud and verbal thought-listing procedures (N ! 88). Half of the children reported their self-talk using both think-aloud and thoughtlisting while the remainder engaged in thought-listing only. Prior participation in think-aloud did not influence the self-talk subsequently reported by children during verbal thought-listing. Compared with thought-listing, more problem-solving (analytical) self-talk and less valenced self-talk was reported during think-aloud. In Study 2, 41 children reported their self-talk via both verbal thought-listing and videomediated recall (own and other’s perspective) procedures. Video-mediated recall generated self-talk of a similar valence to that obtained by thought-listing. More self-talk was generated when the children viewed their own perspective videotape compared with observer perspective videotape and thought-listing. While pre-adolescent children were able to respond to all three cognitive assessment methods, the self-talk produced was not identical. In choosing a cognitive assessment method, researchers and clinicians should be guided by the purpose of the assessment and the setting in which it occurs. KEY WORDS: children; self-talk; think-aloud; thought-listing; video-mediated recall.

Over the last 20 years, interest in the role of cognitions in adult psychopathology has increased dramatically. This has resulted in the development of a number of procedures for assessing cognitions. These procedures differ on four dimensions: structure, endorsement versus free report; timing, concurrent versus retrospective; response mode, written versus verbal; and the nature of the stimulus, in vivo situation, viewed on videotape, or imagined (Glass, 1993). Glass and Merluzzi (1981) classify the available cognitive assessment procedures into six broad categories: University of Otago, Dunedin, New Zealand. Correspondence should be sent to Dr. Gail Tripp, Department of Psychology, University of Otago, P.O. Box 56, Dunedin, New Zealand. E-mail: [email protected]

1 2

399 0147-5916/00/0800-0399$18.00/0  2000 Plenum Publishing Corporation

400

Lodge, Tripp, and Harte

concurrent procedures, recall procedures, prompted recall procedures, recognition/ endorsement approaches, projective methods, and naturalistic approaches. Recent interest in the importance of cognitive factors in child psychopathology has lead to these procedures being used, with few modifications, to assess the cognitions of children. It is unclear, however, which of these techniques is most suitable for use with children, and whether children report different patterns of self-talk in response to the different procedures. The present paper describes two studies that examined the extent to which different cognitive assessment procedures yield comparable self-talk data in preadolescent children. Study 1 compared the self-talk generated by concurrent (thinkaloud) and recall (verbal thought-listing) procedures in a sample of 9- to 10-yearold school children under conditions of mild anxiety. The study also briefly examined the relationship between self-talk type and anxiety. Study 2 compared the self-talk generated by recall (verbal thought-listing) and prompted recall (video-mediated recall) procedures in a sample of 9- to 11-year-old school children. Think-aloud and thought-listing techniques, including video-mediated recall, were selected for study because (1) they come closer than other cognitive assessment methods to meeting Meichenbaum and Cameron’s (1981) guidelines for the collection of valid cognitive self-report data, (2) less is known about the validity of these production methods with children than endorsement procedures, and (3) in the case of think-aloud and thought-listing, both techniques have been compared with endorsement procedures, but to date have not been compared with one another (Houston, Fox, & Forbes, 1984; Kendall & Chansky, 1991; Prins & Hanewald, 1997). STUDY 1 Think-aloud is a concurrent technique in which an individual is asked to verbalize his or her self-talk aloud while simultaneously performing a pre-determined task (Arnkoff & Glass, 1989; Genest & Turk, 1981; Martzke, Andersen, & Cacioppo, 1987). Thought-listing is a recall technique that requires an individual to retrospectively provide (Kendall & Chansky, 1991; Prins, 1985, 1986) verbal or, more usually, written reports of the self-talk experienced during a specific time period (Blackwell, Galassi, Galassi, & Watson, 1985; Halford & Sanders, 1988; Sanders & Dadds, 1992). Think-aloud has the advantage of reducing potential memory and/or reporting bias because the time between experiencing and reporting a self-statement is minimal (Blackwell et al., 1985). Furthermore, it typically yields a large sample of self-talk (Merluzzi & Boltwood, 1989). Thought-listing, on the other hand, is less likely to interfere with task performance (Blackwell et al., 1985). We are aware of only one published study that has directly compared thinkaloud and thought-listing procedures and this was with adults (Blackwell et al., 1985). Think-aloud generated approximately twice as many self-statements as written thought-listing. When both the frequency and proportion of self-talk type were considered, think-aloud was found to produce significantly more attention-control and problem-solving self-statements than thought-listing. Although think-aloud and thought-listing procedures have not been directly compared with children, a small

Assessment of Children’s Self-Talk

401

number of studies have shown that children can respond to these procedures, producing codeable self-talk (Fox, Houston, & Pittner, 1983; Houston et al, 1984; Kendall & Chansky, 1991; Prins, 1985, 1986). In the present study, think-aloud and verbal thought-listing procedures were used to elicit the self-talk of pre-adolescent children placed under conditions of mild anxiety. Verbal thought-listing was used to eliminate the risk that writing limitations would reduce or distort the children’s responses. Data were collected under conditions of mild anxiety to facilitate examination of the relationship between anxiety and self-talk type. The specific aims of the study were to (1) compare the self-talk generated by children using think-aloud and verbal thought-listing procedures, (2) explore whether prior engagement in think-aloud influences the self-talk children subsequently report during thought-listing, and (3) investigate the relationships between self-talk type and anxiety in children. Based on previous research using think-aloud or verbal thought-listing procedures with children, it was assumed that the majority of children would report selftalk (Fox et al., 1983; Houston et al., 1984; Kendall & Chansky, 1991; Prins, 1985, 1986). Previous research with adults (Blackwell et al., 1985) and our own pilot work with children suggested that the children would report more self-talk during thinkaloud than verbal thought-listing. Although Blackwell et al. (1985) found that adults reported more ‘‘problem-solving’’ (analytical) self-statements under think-aloud than thought-listing, this finding was not supported by our preliminary results (Lodge, Harte, & Tripp, 1998). No specific predictions were made about the comparability of self-talk types produced under think-aloud and thought-listing procedures. Because there is currently no research examining the effect of think-aloud on children’s subsequent verbal thought-listing, no hypotheses were formulated regarding this issue. The limited research on the relationship between anxiety and self-talk type in children suggested that negative self-talk would be positively correlated with anxiety (Prins, 1985, 1986; Prins, Groot, & Hanewald, 1994; Zatz & Chassin, 1983, 1985). Method Participants The participants were 88 children (46 females and 42 males) aged between 8 years 8 months and 10 years 9 months (M ! 9 years 8 months). Children 8 years and older were included in the study because reports in the literature, and our own research, suggest that children younger than this cannot reliably report their selftalk. The children were recruited from four suburban primary schools. One school served a low socioeconomic area, two serviced mid socioeconomic areas, and the fourth school serviced a high socioeconomic area. Most (83%) of the participating children were white. Letters of invitation explaining the study were distributed to the parents of children in classes equivalent to third and fourth grade at the four schools. Parents who were willing to allow their child to participate in the study returned a signed consent form to the researchers. Ninety (45.5%) of the 198 consent forms distributed

402

Lodge, Tripp, and Harte

were returned. Participation rates for the four schools ranged from 38.8% (mid socioeconomic area) to 68.2% (high socioeconomic area). Parents who did not return consent forms were not followed up. Materials State-Trait Anxiety Inventory for Children (STAIC). The children’s state- and trait-anxiety were assessed with the STAIC (Spielberger, 1973). The STAIC contains two subscales—a 20-item State scale that assesses the child’s anxiety at the time the inventory is completed, and a 20-item Trait scale designed to measure the child’s generalized feelings of anxiety. The STAIC was designed for use with fourth-, fifth-, and sixth-grade children. Reliability and validity data are available for this measure (Kendall & Ronan, 1990; Spielberger, 1973). The STAIC was administered verbally because previous research indicates New Zealand children may have difficulty with some of the vocabulary used in the scale (Tripp & Lodge, 1993). Math Problems.3 Five math problems, to be completed within 5 minutes, served as the mildly anxiety-provoking task for the participating children. These math problems were taken from the book ‘‘Short Maths 2’’ (Lockyer, 1990). They were selected in consultation with an experienced primary school principal, who considered them to be age appropriate. Given sufficient time, most 8- to 10-yearold children should be able to complete the math problems. Procedure Children with parental permission to participate in the study were approached at school by J.L., who described the study and invited them to take part. All of the children approached consented to participate. The children were individually interviewed twice over a 2-week period (1 week between sessions). Interviews were conducted in a quiet room at the children’s schools. During the first session, the children were verbally administered the trait scale of the STAIC. After this session, pairs of children were matched for age, sex, and STAIC-trait anxiety percentile score. Children within the matched pair were then randomly assigned to one of two groups: think-aloud followed by verbal thoughtlisting (Group 1) or verbal thought-listing only (Group 2). Both groups included 23 girls and 21 boys. The groups did not differ significantly on trait anxiety percentile scores (Group 1, M ! 39, SD ! 26; Group 2, M ! 39, SD ! 26) or age (Group 1, M ! 115.8 months, SD ! 6.9; Group 2, M ! 116.7 months, SD ! 5.9). At the beginning of the second session, the children were informed that they would be given 5 minutes in which to complete a series of math problems. They were told that the researcher was interested in how many of the problems they could get right within this time and that, although they might not get all of the problems finished, they should try their best to work them out. After a short practice trial, the children in Group 1 were given 2 minutes to report aloud everything that they were thinking and saying to themselves (think-aloud). Children who remained silent for 10 seconds were given a maximum of two standard prompts: ‘‘Remember Copies of the five math problems are available from the second author.

3

Assessment of Children’s Self-Talk

403

to say out loud all the things that come into your head’’ and ‘‘What are you thinking and saying to yourself right now?’’ Children in both groups were then verbally administered the STAIC-state scale. Following this, all children were given the math problems to work on. Children in Group 1 were asked to say out loud everything they were thinking and saying to themselves while they worked on the problems (think-aloud). As before, children who remained silent for 10 seconds were prompted (maximum of two standard prompts). The researcher also responded briefly to occasional questions raised by some children in both groups. No other verbal contact was made with the children while they worked on the math problems. The researcher was positioned at the back of the room, within hearing distance but outside the children’s line of vision, during this 5-minute period. Self-talk generated prior to and during the maths problems was combined for analysis. After 5 minutes, the children were asked to stop working on the problems and were interviewed about what they were thinking prior to and while participating in the task (verbal thought-listing). The children were requested to: 1. ‘‘Tell me everything you thinking and saying to yourself when I first told you about the math problems?’’ 2. ‘‘Tell me everything you thinking and saying to yourself while you were working on the math problems?’’ Standard prompts were given if children failed to respond within 10 seconds (‘‘Were you thinking or saying anything to yourself?’’ If an affirmative response was given the child was asked ‘‘what were you thinking or saying to yourself?’’) or responded that they were thinking about the situation rather than reporting their actual selftalk (‘‘What were the thoughts that went through your mind?’’). The children were asked to indicate when they had finished responding to each thought-listing ‘‘question’’ by raising a hand. The children’s self-talk during think aloud and/or thought-listing was audiotaped and later transcribed verbatim. Results Parametric and nonparametric tests were used to analyze the data. Nonparametric procedures were used where data were categorical in nature or non-normally distributed. Unless otherwise stated, critical alpha was set at .05 and analyses were two-tailed. Coding of Dependent Variables The transcripts of the children’s self-talk were unitized by two trained raters working independently. A unit was defined as representing one main idea, rather than meaningless words. Sentence structure and content were used to help define units. Copies of the unitizing manual are available from the second author. One rater unitized all of the transcripts, while the second rater unitized self-statements from randomly selected pages within the transcripts (45% of the transcribed material). Throughout unitizing, the raters had access to the audiotapes of the children’s self-talk.

404

Lodge, Tripp, and Harte

Agreement over the existence of a ‘‘unit’’ was defined as the two raters marking ‘‘unit boundaries’’ in exactly the same location on the transcripts. The two raters agreed on 97.1% of the commonly unitised material (372 of the 383 units identified). There were 11 omission disagreements (i.e., ‘‘unit boundaries’’ that only one of the two raters identified). Disagreements were resolved in favor of the rater who unitized all of the transcripts. Following unitizing, the units of self-talk were coded into six theoretically meaningful categories: positive, negative, neutral, analytical, directive (sometimes referred to as coping), and questioning self-talk using a system developed by J.L. Questions are not usually identified separately but coded according to valence. A category of questioning self-talk was included in light of the frequency with which questions appear in the self-talk of anxious adults (Kendall & Ingram, 1987). Descriptions of all of the self-talk categories, together with examples, are given in Table I. Table I. Cognitive Content Categories: Definitions and Examples Cognitive category Positive

a

Negativea

Neutrala

Analytical

Directivea,b

Questioning

Definitions Self-statement(s) likely to facilitate control of fear/anxiety and/or achieving a favorable outcome on (or completion of ) the task. Excludes questions (coded as Questioning) and self-instructions (coded as Directive). Self-statements likely to hinder control of fear and/or hinder a favorable outcome on (or completion of ) the task. Excludes questions (coded as Questioning). Self-statements that have no clear influence on the task at hand and cannot be clearly categorized elsewhere. On-task verbalizations in which the subject is engaging in problem-solving strategies to complete the task. Self-statements directing thought or behavior, expressed in the form of a self-instruction. Unlikely to interfere with achieving a favorable outcome on the task, task completion, or controlling fear/anxiety. Self-statements expressed in the form of a question. Excludes reading questions aloud from the math task or asking questions in order to solve a math problem (e.g., ‘‘What comes after 10?’’) (both coded as Analytical).

Examples ‘‘I can do it.’’ ‘‘I enjoy math.’’ ‘‘This is a challenge.’’

‘‘This looks hard.’’ ‘‘I must be stupid.’’ ‘‘I’m getting them all wrong.’’ ‘‘I’m thinking about my best friend.’’ ‘‘There are five questions.’’ ‘‘This is like being in the classroom.’’ ‘‘The answer is 10.’’ ‘‘A B C D E.’’ ‘‘I’ll work this out by adding up the numbers.’’ ‘‘Work at my own pace.’’ ‘‘Stay calm.’’ ‘‘I must keep concentrating.’’

‘‘What if I can’t finish these in time?’’ ‘‘Will I get it done?’’ ‘‘Wonder how I’m doing?’’

Adapted from Lodge et al., 1998. a Definitions based on Kendall & Chansky (1991). b ‘‘Directive’’ self-talk is more commonly referred to as ‘‘coping’’ self-talk in the literature.

Assessment of Children’s Self-Talk

405

A further two raters, blind to the experimental hypotheses, independently coded the units of self-talk. Each rater received approximately 10 hours of training in the use of the coding system and began coding the unitised transcripts once they reached more than 95% agreement with J.L. One rater categorized all units on the transcripts, while the other categorised 388 (47.5%) of the 817 units from randomly selected pages within the transcripts. Approximately 7% of the units were not coded (e.g., acknowledgements of prompts, descriptions of behavior, questions directed at the interviewer). Inter-rater agreement was calculated using Cohen’s Kappa (Bakeman & Gottman, 1986; Cohen, 1960). Kappas for the individual self-talk categories ranged from 0.87 to 1.00, with an overall kappa of 0.96. Disagreements in coding were resolved in favor of the rater who coded all of the transcribed units. Anxiety Levels The participants’ raw scores on the state- and trait-scales of the STAIC were converted to percentile scores. Paired t tests were used to compare state and trait percentile scores within each group. Children in both groups reported experiencing significantly more anxiety just prior to completing the math problems than in general (Group 1: State M ! 53, Trait M ! 39, t[43] ! 2.89, p " .01; Group 2: State M ! 55, Trait M ! 39, t[43] ! 4.09, p " .001). Nature of Self-Talk Reported During Think-Aloud and Thought-Listing All of the children reported self-talk during think-aloud (Group 1). One child in Group 1 and six children in Group 2 reported no self-talk during thought-listing. Chi-square analysis showed that the number of children failing to report self-talk during thought-listing did not differ significantly across the groups. Table II presents the range, mean, and modal number of self-statements reported by each group for each cognitive assessment method. The median amount of self-talk generated during think-aloud and thoughtlisting procedures by the children in Group 1 was compared with a Wilcoxon matched-pairs signed-ranks test. Significantly more self-talk was reported during think-aloud (M ! 9.5, SD ! 6.8) than during thought-listing (M ! 4.6, SD ! 2.7) Z ! #4.91, p " .0001.4 To compare the type of self-talk generated the responses in each self-talk category were converted to proportions of the total amount produced. This was done to control for differences in the amount of self-talk generated across methods. Only data from children who actually reported self-talk were converted to proportions. It is not meaningful to talk about the proportions of self-talk types for children who reported no self-talk. These proportion data are presented in Table III. The proportion data show a similar pattern to the frequency data. For thinkaloud, analytical self-talk was most commonly reported, followed by negative and questioning self-talk. For thought-listing, negative and positive self-talk were the This analysis was repeated removing the data from the child who reported no self-talk during thoughtlisting: think aloud M ! 9.6, SD ! 6.8; thought-listing M ! 4.7, SD ! 2.7; Z ! #4.84, p " .0001, N ! 43.

4

406

Lodge, Tripp, and Harte

Table II. Frequency of Self-Statements Assigned to Each Cognitive Category According to Group and Cognitive Assessment Method

Median

Range

No. (%) reporting cognition type

N ! 44

N ! 44

N ! 44

N ! 44

1.8 (2.1) 1.5 (1.9) 3.4 (3.9) 1.6 (1.8) 1.0 (1.6) 0.2 (0.5) 9.5 (6.8) N ! 44 1.4 (1.4) 1.2 (1.3) 0.7 (1.1) 0.8 (1.3) 0.4 (0.8) 0.1 (0.2) 4.6 (2.7) N ! 44 0.9 (1.2) 0.8 (1.2) 0.7 (0.9) 0.5 (1.0) 0.2 (0.5) 0.1 (0.3) 3.2 (2.2)

1.0 1.0 2.5 1.0 0.0 0.0 8.0 N ! 43 1.0 1.0 0.0 0.0 0.0 0.0 4.0 N ! 38 1.0 1.0 1.0 0.0 0.0 0.0 3.5

0–8 0–7 0–15 0–7 0–6 0–2 2–31 N ! 43 0–6 0–6 0–4 0–6 0–3 0–2 1–12 N ! 38 0–5 0–5 0–3 0–3 0–2 0–1 1–8

Number of self-statements reported

Cognition category

Mean

Group 1 (TA) Negative Positive Analytical Questioning Neutral Directive Total Group 1 (TL)a Negative Positive Analytical Questioning Neutral Directive Total Group 2 (TL)a Negative Positive Analytical Questioning Neutral Directive Total

(S.D.)

N ! 43 1.4 (1.4) 1.3 (1.3) 0.7 (1.1) 0.8 (1.3) 0.4 (0.8) 0.0 (0.2) 4.7 (2.7) N ! 38 1.0 (1.2) 0.9 (1.3) 0.8 (0.9) 0.6 (1.0) 0.2 (0.5) 0.1 (0.3) 3.7 (1.9)

29 25 30 27 18 8

(65.9%) (56.8%) (68.2%) (61.4%) (40.9%) (18.2%)

N ! 43 30 (69.8%) 32 (74.4%) 17 (39.5%) 15 (34.9%) 12 (27.9%) 2 (4.7%) N ! 38 22 (57.9%) 20 (52.6%) 21 (55.3%) 10 (26.3%) 6 (15.8%) 3 (7.9%)

TA ! think-aloud; TL ! verbal thought-listing. a TL means and standard deviations using the entire sample and only those children reporting self-talk are presented.

most prevalent, followed by high rates of analytical self-talk for children in Group 2. Because the majority of the self-talk was assigned to the negative, positive, analytical, and questioning categories, only these four types are considered further. The proportion of each self-talk type (negative, positive, analytical, and questioning) reported during think-aloud and thought-listing was compared with Wilcoxon matched-pairs signed-ranks tests. Proportions of negative (Z ! #2.59, p " .01), positive (Z ! #3.08, p " .005), and analytical (Z ! #3.63, p " .0005) self-talk, but not questioning self-talk, differed significantly across the cognitive assessment methods. Specifically, higher proportions of negative and positive self-talk were reported during thought-listing, whereas a higher proportion of analytical self-talk was reported during think-aloud. Consideration of the frequency data suggest that the differences in the proportions of positive and negative self-talk observed between think-aloud and thoughtlisting are an artifact of the very high rate of analytical self-talk reported during think-aloud. The think-aloud/thought-listing analyses were repeated using the frequency data (N ! 43). Wilcoxon matched-pairs signed-rank tests indicated that the number of analytical (Z ! #4.37, p " .0001) and questioning (Z ! #2.68, p "

Assessment of Children’s Self-Talk

407

Table III. Proportion of Self-Statements Assigned to Each Cognitive Category According to Group and Cognitive Assessment Method Cognition category Group 1 (TA) Negative Positive Analytical Questioning Neutral Directive Group 1 (TL) Negative Positive Analytical Questioning Neutral Directive Group 2 (TL) Negative Positive Analytical Questioning Neutral Directive

Proportion of self-statements reported Meana (S.D.)

Median

Range

0.22 0.15 0.33 0.21 0.08 0.01

(0.26) (0.19) (0.30) (0.28) (0.12) (0.03)

0.13 0.09 0.35 0.13 0.00 0.00

0.00–1.00 0.00–0.88 0.00–1.00 0.00–1.00 0.00–0.50 0.00–0.13

0.34 0.27 0.13 0.17 0.08 0.01

(0.31) (0.24) (0.19) (0.27) (0.14) (0.05)

0.29 0.25 0.00 0.00 0.00 0.00

0.00–1.00 0.00–1.00 0.00–0.67 0.00–1.00 0.00–0.50 0.00–0.33

0.27 0.24 0.24 0.15 0.07 0.03

(0.30) (0.30) (0.28) (0.28) (0.19) (0.30)

0.23 0.17 0.21 0.00 0.00 0.00

0.00–1.00 0.00–1.00 0.00–1.00 0.00–1.00 0.00–1.00 0.00–0.33

TA ! think-aloud; TL ! verbal thought-listing. a Means are based on the number of children reporting selftalk (Group 1/TA ! 44; Group 1/TL ! 43; Group 2/TL ! 38)

.01) self-statements reported differed significantly across methods. More analytical and questioning self-talk were reported during think-aloud than during thoughtlisting. Frequencies of positive and negative self-talk did not differ across cognitive assessment procedures. Effect of Think-Aloud on Subsequent Thought-Listing To determine whether prior participation in think-aloud influenced the selftalk reported during thought-listing, both the total amount of self-talk and the proportions of self-talk types (negative, positive, analytical, and questioning) reported during thought-listing were compared across the groups using Mann-Whitney U tests. Children who reported no self-talk were excluded from the analyses of self-talk type. When only those children who reported self-talk were included in the analysis, there was no significant group difference in the total amount of self-talk generated during thought-listing. However, when the analysis was repeated to include all of the children, the group who had engaged in think-aloud prior to thought-listing (M ! 4.5, SD ! 2.7) reported significantly more self-statements than those who engaged in thought-listing only (M ! 3.2, SD ! 2.2) Z ! #2.3, p " .05. No significant group differences were found for the proportions of self-talk types reported, although the difference in proportion of analytical self-talk reported

408

Lodge, Tripp, and Harte

Table IV. Spearman Rank Correlations Between Anxiety (STAIC-State and Trait Percentile Scores) and Self-Talk Type Think-aloud (N ! 44) STAIC-trait

STAIC-state

Thought-listing (N ! 81) STAIC-trait

STAIC-state

Frequency of self-talk Negativea Positive Questioninga Analytical

rs ! #.16 ns rs ! .11 ns rs ! .04 ns rs ! #.05 ns

rs ! .15 ns rs ! #.08 ns rs ! #.18 ns rs ! #.21 ns

rs ! rs ! rs ! rs !

.14 ns .04 ns .01 ns .03 ns

rs ! .37** rs ! #.08 ns rs ! #.13 ns rs ! .09 ns

Proportion of self-talk Negativea Positive Questioninga Analytical

rs ! #.15 ns rs ! .06 ns rs ! #.03 ns rs ! #.05 ns

rs ! .25* rs ! .01 ns rs ! #.13 ns rs ! #.25 ns

rs ! .10 ns rs ! #.01 ns rs ! .02 ns rs ! .00 ns

rs ! .35** rs ! #.13 ns rs ! #.12 ns rs ! .06 ns

* p " .05. ** p " .0005 ns ! not significant. a Where the direction of a relationship was predicted, one-tailed tests were used.

approached significance (Z ! #1.95, p ! .052).5,6 Children in Group 2 reported a higher proportion of analytical self-talk than those in Group 1. Self-Talk and Anxiety Spearman rank correlations were used to assess the degree of linear association between both the frequency and proportion of positive, negative, questioning, and analytical self-talk and the children’s levels of anxiety (Table IV). Separate correlations were carried out for think-aloud and thought-listing, as the self-talk generated by these cognitive assessment procedures differed. Because there were no significant group differences in the self-talk types reported during thought-listing, these data were collapsed across the groups. Alpha was set at .01 to protect against type I errors. For thought-listing, both the proportion (rs ! .35, p " .0005, one-tailed) and the frequency (rs ! .37, p " .0005, one-tailed) of negative self-talk were significantly positively correlated with state anxiety levels. For think-aloud, the correlation between state anxiety and the proportion of negative self-talk (rs ! .25, p ! .026, one-tailed) approached, but did not reach, statistical significance. All other correlations were nonsignificant. Discussion The results of the present study indicate that the self-talk reported by children is influenced by the cognitive assessment method used. Think-aloud and thoughtlisting procedures do not produce identical self-talk. Although prior participation These analyses were repeated to include the children who reported no self-talk during verbal thoughtlisting. Mann-Whitney U tests showed no significant group differences in the proportions of negative, positive, analytical, and questioning self-talk reported. 6 Any tests that approached statistical significance were checked using randomization tests (Manly, 1992). Results were consistent with the non-parametric analyses reported. 5

Assessment of Children’s Self-Talk

409

in think-aloud does not influence the nature of the self-statements subsequently reported during thought-listing, there is some evidence to suggest it may facilitate the generation of self-talk via thought-listing. Consistent with previous research with adults (Blackwell et al., 1985), thinkaloud yielded more self-talk than thought-listing. The difference in the amount of self-talk reported may reflect the children’s increased access to task-related cognitions during think-aloud. This idea is supported by the greater proportion of analytical self-talk generated during think-aloud. Think-aloud is assumed to access selftalk from short-term memory, whereas thought-listing is believed to tap self-talk from longer-term memory. Difficulty in retrieving self-talk from long-term memory, together with potential interference from new thoughts occurring during the recall process, may lead to omissions in or distortions of the self-talk generated during thought-listing. This may result in fewer self-statements being reported under this procedure (Blackwell et al., 1985). Alternatively, the children may have simply had more opportunity to report their self-talk during think-aloud. During this procedure, the tape recorder was left running for 2 minutes prior to the task and throughout task participation. Under thought-listing the children could continue to report their self-talk for as long as they wished. Most did not take this opportunity, however. In addition to the amount of self-talk reported, the two assessment procedures yielded different patterns of results with respect to self-talk type. Think-aloud produced significantly more analytical self-talk than thought-listing. The difference in analytical self-talk generated by these procedures can also be explained in terms of access to cognitions. The children were asked to think aloud while they were actually solving the math problems, a time when they were likely to be engaging in analytical self-talk. When questioned about the self-talk they had engaged in while working on the math problems (i.e., thought-listing), the children may have reported their overall problem-solving strategy rather than retracing the steps used to solve the problems. Proportionally more positive and negative self-talk was generated by thoughtlisting than think-aloud. This suggests that, after adjusting for variability in the amount of self-talk reported, thought-listing may be a useful procedure for accessing cognition valence. On the other hand, this finding may be an artifact of the high rate of analytical self-talk reported during think-aloud. Interestingly, we did not find that thought-listing resulted in the bias toward negative cognition that other researchers have identified (Kendall & Chansky, 1991; Prins & Hanewald, 1997). Proportion data can clearly produce a different pattern of results from that obtained with frequency data. The decision as to whether to give greater weight to frequency data or proportion data is not an easy one; both have strengths and limitations. Proportion data takes account of variability in the report of self-talk across individuals and procedures whereas frequency data does not. However, with frequency data, the rate of any one type of self-talk is not dependent on the rate of any other type of self-talk. As the pattern of self-talk reported via think-aloud and thought-listing differed, the relationships between self-talk types and anxiety were examined separately for the two cognitive assessment methods. It was predicted that increased anxiety would be associated with higher rates of negative self-talk. The present results support

410

Lodge, Tripp, and Harte

this prediction for state, but not trait anxiety. As the sample was not generally anxious, a relationship between trait anxiety and negative self-talk might not be expected. The relationship between negative self-talk and state anxiety was stronger for thought-listing than for think-aloud. This pattern of results is not altogether surprising, given the suggestion that thought-listing may be more effective than think-aloud in accessing cognition valence. Alternatively, the stronger correlation found for thought-listing may simply reflect the larger sample size.

STUDY 2 Video-mediated recall is a prompted recall approach in which an individual is videotaped while engaging in the target behavior. The videotape is later played back to the individual, who is asked to reconstruct or ‘‘dub’’ the self-talk engaged in during the target behavior, using the videotape to aid recall. It has been suggested that providing retrieval cues to aid the recall of self-talk may result in more reliable and complete recall than would be the case with thought-listing (Martzke et al., 1987). Other researchers have questioned whether the dubbed self-talk represents a post hoc account of the behavior observed on videotape (Merluzzi & Boltwood, 1989). To date, only one study has used video-mediated recall with children (Sanders & Dadds, 1992). The results of this study suggest that video-mediated recall yields a different pattern of self-talk than that generated by written thought-listing. In discussing their findings, the authors suggest these differences may be due, at least in part, to the difficulty children have writing rather than verbalizing their self-talk. In view of this, the present study adopted a verbal thought-listing procedure. Questions have been raised about the importance of camera position in the use of video-mediated recall. Schwartz and Garamoni (1986) suggest that videotaping an individual from an observer’s perspective may result in increased self-awareness and a subsequent increase in negative self-talk. They consider that positioning the camera from the participant’s perspective provides a more effective retrieval cue because it gives a more accurate representation of the individual’s experience. Although a seemingly sound suggestion, Schwartz and Garamoni’s (1986) hypothesis remains to be tested. In the present study, pre-adolescent children were videotaped from two different camera positions while completing the previously described math problems and a puzzle task under time pressure. The children reported their self-talk via verbal thought-listing immediately after completing the two tasks and subsequently in response to viewing their videotapes. The aims of the study were to (1) compare the amount and nature of selftalk generated by children using verbal thought-listing and video-mediated recall procedures, and (2) explore the influence of camera position on the self-talk generated. Given the exploratory nature of this study, no specific hypotheses are offered.

Assessment of Children’s Self-Talk

411

Method Participants The participants were 41 children (27 females and 14 males) ranging in age from 9 years to 11 years 1 month (M ! 10 years 2 months). They were drawn from two local primary schools, servicing mid socioeconomic areas. The majority (90%) of the participants were white. Recruitment procedures were identical to those used in Study 1. Forty-one (54%) of the 76 consent forms distributed were returned. Parents who did not return consent forms were not followed up. Complete data sets were available for 35 of the 41 children. Materials In addition to the materials used in Study 1, the present study included a second, mildly anxiety-provoking task referred to as ‘‘the puzzle.’’ Furthermore, this study used two video cameras to record the children working on the two tasks. A video player and color monitor were used to play the videotapes back to the participants. Puzzle. Design nine from the block design subtest of the Wechsler Adult Intelligence Scale—Revised served as the puzzle (Wechsler, 1981). This task required the children to arrange nine colored blocks to match the design shown on a colored card. Although designed for completion by adults, a small number of children in the current study completed the puzzle in the 5 minutes allocated. Procedure At the request of the schools, D.K.H. spent a morning in the children’s classrooms prior to beginning data collection. Following this, children with parental permission to participate were approached in class by D.K.H. and asked if they wanted to take part. All children approached consented to participate. The children were individually interviewed three times over 3 successive days (one session per day). Interviews were conducted in a quiet room at the children’s schools. At the beginning of the first session, the children were told they would be required to complete two tasks (a series of math problems and a puzzle) within a limited time (5 minutes per task). The children were informed that, although they might not get the problems/puzzle finished, they should try their best to complete the tasks. Before beginning the tasks, but after being told what was expected of them, the children were verbally administered the state scale of the STAIC. They then attempted the math problems and the puzzle, with the order of task presentation counterbalanced across pairs of children (matched for age and sex). The children’s participation in the two tasks was concurrently videotaped from both an observer’s perspective (video camera in front of the child) and the child’s own perspective (video camera behind the child, focusing on the desk top) for use in the two subsequent sessions. The researcher remained in the room, within hearing range but outside the children’s vision, during this time. The only verbal contact she had with the children was to respond to their occasional questions.

412

Lodge, Tripp, and Harte

After working on each task, the children were asked a series of questions about their thoughts and feelings prior to, while working on, and immediately after task participation. The children were asked: 1. ‘‘What were you thinking and saying to yourself/how were you feeling just before I gave you the math problems/puzzle?’’ 2. ‘‘What were you thinking and saying to yourself/how were you feeling while you were working on the math problems/puzzle?’’ 3. ‘‘What were you thinking and saying to yourself/how were you feeling as soon as you finished doing the math problems/puzzle?’’ Questions about thoughts and feelings were counterbalanced within the matched pair. Children who failed to respond within 10 seconds of a question were given the standard prompts used in Study 1. On completion of the two tasks and associated questions, the children were verbally administered the trait scale of the STAIC. The videotapes of the children completing the math and puzzle tasks were edited into three 30-second segments. The first segment showed the children receiving instructions prior to beginning the task; the second showed the children working on the task half-way through the 5 minute time period; and the third showed the children working on the task in the last 30 seconds of the available time. There were four edited videotapes per child—one for each task and camera angle. During the second and third sessions, the children watched their videotapes of task participation. In each session, half of the children watched their ‘‘own perspective’’ videotapes for both the math and puzzle tasks, while the remainder of the children watched their ‘‘observer’s perspective’’ videotapes. The order of task presentation was the same as the order in which the children completed the tasks. The children were instructed to watch the videotapes closely and to try to recall everything they had been thinking and feeling during the period depicted. All video segments were played without sound so that the children would not be influenced by any verbalizations they had made while working on the tasks. At the end of each 30-second segment the children were asked about the thoughts and feelings they had experienced in the scene viewed. As with Session 1, questions about thoughts and feelings were counterbalanced across pairs of participants. At the end of the third session, the children were asked about which video perspective (own or observer) they preferred watching. All self-talk was audiotaped for later transcription and coding. As children often reported self-talk in response to questions about feelings, these responses were included in all analyses of children’s self-talk. Results Coding of Dependent Variables As in Study 1, the transcripts of the children’s self-talk were unitized by two trained raters working independently. One rater unitized all of the transcripts, while

Assessment of Children’s Self-Talk

413

the second rater unitised 31.6% of the transcribed material. The two raters agreed on 97.4% of the commonly unitised material (594 of the 610 possible units). Disagreements were resolved in favor of the rater who unitised all of the transcripts. The unitized transcripts were then coded by a second pair of trained raters, blind to the study’s purpose. The coding system described in Study 1 was also used here. One rater coded all 1889 units on the transcripts, while the second rater coded 687 (36.4%) units. All coding disagreements were resolved in favor of the rater who coded all of the transcripts. Cohen’s kappa (Bakeman & Gottman, 1986; Cohen, 1960) for the individual categories of self-talk ranged from 0.84 to 1.00, with an overall kappa of 0.94. Throughout unitizing and coding, the raters had access to the audiotapes of the children’s self-talk. In this study approximately 13% of the units were not coded. Total Self-Talk Independent t tests were conducted, separately for each camera angle (own perspective; observer’s perspective), to determine whether the order in which these videotapes were presented (first, second) influenced the total amount of self-talk reported during video-mediated recall. No order effects were found and order of videotape presentation was excluded from subsequent analyses. The effect of condition (thought-listing, observer’s perspective, own perspective) and task type (math, puzzle) on the total amount of self-talk reported was tested with a 3 (condition) $ 2 (task type) repeated measures MANOVA.7 There was no significant effect for task type, nor condition by task type interaction; however, there was a significant main effect for condition, F [2,33] ! 4.35, p " .021 (Phillis). Univariate F tests indicated there was no difference between the total self-talk reported under thought-listing and observer’s perspective conditions, but a significant difference between the average of the self-talk reported under these two conditions and the own perspective, F [1,34] ! 8.94, p ! .005. More self-talk was generated under the own perspective condition than either thought-listing or observers perspective. As the assumption of sphericity of the variance-covariance matrix was violated the univariate results were corrected using both the GreenhouseGeisser epsilon (F [1,26] ! 8.94, p " .02) and the Huynh-Feldt epsilon (F [1,27] ! 8.94, p " .02) (Norusis & SPSS Inc., 1994). Self-Talk Valence To control for method variance in the amount of self-talk reported, responses in each of the six cognitive content categories were converted to proportions of the total amount of self-talk reported. These proportion data are presented in Table V. As the majority of self-talk reported across tasks and methods was either positive or negative, only these two types of self-talk are considered further. Two separate 3 (condition) $ 2 (task type) repeated measures MANOVAs7 were carried out to determine if the proportions of negative and positive self-talk Analysis conducted on the data from the 35 children for whom complete data sets were available.

7

414

Lodge, Tripp, and Harte

Table V. Proportions of Positive and Negative Self-Statements Reported During Thought-Listing and Video-Mediated Recall: Math and Puzzle Tasks Cognitive assessment method & self-talk type

Math task Mean (S.D.)

Puzzle task Mean (S.D.)

Median

Range

(0.28) (0.22) (0.11) (0.14) (0.12) (0.04)

0.33 0.33 0.00 0.00 0.00 0.00

0.11–1.00 0.00–1.00 0.00–0.42 0.00–0.40 0.00–0.50 0.00–0.14

0.29 0.41 0.09 0.10 0.07 0.03

VMR—observer perspective Negative 0.32 (0.27) Positive 0.42 (0.32) Analytical 0.07 (0.10) Questioning 0.10 (0.14) Neutral 0.08 (0.12) Directive 0.01 (0.06)

0.30 0.38 0.00 0.00 0.00 0.00

0.00–1.00 0.00–1.00 0.00–0.33 0.00–0.50 0.00–0.50 0.00–0.33

VMR—own perspective Negative 0.33 Positive 0.38 Analytical 0.09 Questioning 0.06 Neutral 0.13 Directive 0.01

0.25 0.40 0.00 0.00 0.11 0.00

0.00–1.00 0.00–0.83 0.00–0.33 0.00–0.33 0.00–0.50 0.00–0.17

Thought-listing Negative Positive Analytical Questioning Neutral Directive

0.36 0.35 0.08 0.10 0.09 0.02

(0.25) (0.42) (0.11) (0.10) (0.14) (0.03)

Median

Range

(0.21) (0.25) (0.13) (0.14) (0.11) (0.08)

0.25 0.40 0.00 0.00 0.00 0.00

0.00–0.80 0.00–1.00 0.00–0.50 0.00–0.56 0.00–0.50 0.00–0.33

0.36 0.43 0.06 0.11 0.03 0.01

(0.28) (0.28) (0.10) (0.15) (0.09) (0.03)

0.33 0.50 0.00 0.00 0.00 0.00

0.00–1.00 0.00–1.00 0.00–0.33 0.00–0.50 0.00–0.33 0.00–0.17

0.32 0.37 0.08 0.09 0.11 0.03

(0.28) (0.26) (0.13) (0.13) (0.16) (0.08)

0.27 0.33 0.00 0.00 0.00 0.00

0.00–1.00 0.00–1.00 0.00–0.57 0.00–0.36 0.00–0.67 0.00–0.33

Means and percentages are based on the total number of children reporting self-talk for each condition and task (thought-listing math n ! 39, puzzle n ! 38; VMR-Observer Perspective math n ! 39, puzzle n ! 38; VMR-Own Perspective math n ! 39, puzzle n ! 38). VMR ! video-mediated recall.

were influenced by condition (thought-listing, observer’s perspective, own perspective) and task type (math, puzzle). Prior to conducting these analyses, the proportion data were transformed with an arcsine transformation to correct for non-normal distributions (Zar, 1984). For both positive and negative self-talk, there were no significant main effects of condition or task type, nor any significant condition by task type interactions. The relationship between self-talk type and anxiety is discussed separately in Lodge et al. (1998). Preferred Videotape The children were asked which videotape they preferred watching, own or observer’s perspective. Sixty-one percent (N ! 23)of the children indicated they preferred the observer-perspective videotapes, 29% (N ! 11) the own-perspective videotapes, and 10% (N ! 4) had no preference. Using the binomial test, the two-tailed probability of 11 or fewer children preferring the observer perspective videotape approached significance at p ! .058. A chi-square analysis showed there was no effect of gender on videotape preference.

Assessment of Children’s Self-Talk

415

Discussion The results of the second study demonstrate that video-mediated recall generates self-talk of similar valence to that obtained with verbal thought-listing. The proportions of positive and negative self-talk reported during verbal thought-listing did not differ significantly from those generated with own- and other-perspective video-mediated recall. Videotape perspective does not appear to influence the proportions of positive and negative self-talk generated. Similar proportions of both types of self-talk were generated by own and observer’s perspective. This finding contradicts Schwartz and Garamoni’s (1986) suggestion that videotaping from an observer’s perspective may increase self-awareness leading to an increase in negative self-talk. If videomediated recall is used to determine the valence of children’s self talk, then camera position does not appear to be an issue. Placement of the camera may be determined by the demands of the situation. Camera position does, however, influence the amount of self-talk generated. Own perspective video-mediated recall generated significantly more self-talk than either observer’s perspective video-mediated recall or thought-listing. The difference in self-talk generated between thought-listing and own perspective videomediated recall could be an artifact of method. Under video-mediated recall the children actually had more opportunity to verbalize their thoughts. However, this does not explain why more self-talk was generated under own versus observer’s perspective video-mediated recall. The own-perspective videotape may, as suggested by Schwartz and Garamoni (1986), provide a more effective retrieval cue, reflecting more accurately the child’s experience of the math problems and puzzle task. Given that children in the study preferred the observer’s videotape, and that valence is not influenced by perspective, then it would seem sensible to videotape children from the observer’s perspective. If, however, the purpose is to generate as much self-talk as possible, an argument can be made for videotaping from the child’s perspective. Researchers and clinicians who propose to use video-mediated recall will need to weigh the child’s comfort against their need for large samples of self-talk. Although video-mediated recall produces self-talk of similar valence to verbal thought-listing, it requires specialized equipment and is time consuming. It is probably most cost-effective to have children report their self-talk using a thought-listing approach within a short time of completing an activity. Where this is not possible, video-mediated recall offers a viable alternative.

GENERAL DISCUSSION The present studies are among the first to evaluate children’s responses to a variety of cognitive assessment methods. The results indicate that children as young as 9 years of age are able to respond to think-aloud, verbal thought-listing, and video-mediated recall procedures, producing self-talk that is codeable.

416

Lodge, Tripp, and Harte

The three cognitive assessment methods do not produce identical self-talk data. Both think-aloud and video-mediated recall (own perspective) elicit more self-talk than verbal thought-listing. It seems likely that think-aloud and the own-perspective video-mediated recall allowed the children more direct access to their thoughts, resulting in a greater amount of self-talk. Self-talk valence is similar for video-mediated recall and thought-listing, suggesting that engaging in video-mediated recall after thought-listing does not alter the valence of the self-talk initially reported. The valence of self-talk generated by thought-listing following think-aloud is the same as that produced by thought-listing only. However, under think-aloud the proportion of negative and positive self-talk drops, probably reflecting the children’s greater number of task-related cognitions. The results from Study 1 strongly suggest that reporting self-talk with one technique does not alter the nature of the self-talk subsequently reported via another technique. The Study 2 findings also offer tentative support for this proposition. These findings are encouraging, suggesting that researchers or clinicians can safely use more than one assessment method to elicit children’s self-talk. The use of additional methods is unlikely to distort the pattern of self-talk produced, and may facilitate its production. Having identified that children are able to respond to these different assessment methods, reporting self-talk that is codeable, we encourage researchers and clinicians to consider incorporating them into their research and clinical practice. The choice of assessment method will clearly depend on the purpose of the assessment, the setting, and the resources available. Think-aloud generates more task-focused self-talk but is not appropriate in all circumstances. For example, it is not possible to think aloud during a task that requires verbalization. Thought-listing, if carried out shortly after task completion, appears to be a practical alternative, but is likely to produce fewer self-statements. When there is a delay between the task or event and thought-listing, video-mediated recall might be considered. Video-mediated recall is obviously more involved in terms of time and equipment than the other procedures. While acknowledging that production techniques such as those used in the present studies can be labor intensive, they do not restrict an individual to a predetermined set of self-statements as in the case of endorsement approaches. Interestingly, the children’s pattern of responses to the different cognitive assessment methods are very similar to those reported with adults, suggesting a consistency of method effects across a range of ages. For example, adults also report more self-talk and more ‘‘problem-solving’’ statements under think-aloud compared with thought-listing procedures (Blackwell et al., 1985). While providing valuable information on children’s responses to different cognitive assessment methods, the current studies are not without their limitations. In Study 2, video-mediated recall always followed thought-listing. To be certain that the valence of video-mediated recall is not affected by prior participation in thoughtlisting, it would be necessary to counterbalance the techniques. Alternatively, a similar design to Study 1 could be used to test the effect of prior thought-listing on subsequent video-mediated recall. Furthermore, both studies restricted their focus to the amount and valence of the self-talk generated by the different assessment methods. No attempt has been

Assessment of Children’s Self-Talk

417

made to examine the extent to which the content of the self-talk produced overlaps. This is an obvious next step requiring the development of appropriate coding schemes.

REFERENCES Arnkoff, D. B., & Glass, C. R. (1989). Cognitive assessment in social anxiety and social phobia. Clinical Psychology Review, 9, 61–74. Bakeman, R., & Gottman, J. M. (1986). Observing interaction: An introduction to sequential analysis. New York: Cambridge University Press. Blackwell, R. T., Galassi, J. P., Galassi, M. D., & Watson, T. E. (1985). Are cognitive assessment methods equal? A comparison of think aloud and thought listing. Cognitive Therapy and Research, 9(4), 399–413. Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1), 37–46. Fox, J. E., Houston, B. K., & Pittner, M. S. (1983). Trait anxiety and children’s cognitive behaviors in an evaluative situation. Cognitive Therapy and Research, 7(2), 149–154. Genest, M., & Turk, D. C. (1981). Think-aloud approaches to cognitive assessment. In T. V. Merluzzi, C. R. Glass, & M. Genest (Eds.), Cognitive assessment. New York: Guilford. Glass, C. R. (1993). A little more about cognitive assessment. Journal of Counseling and Development, 71, 546–548. Glass, C. R., & Merluzzi, T. V. (1981). Cognitive assessment of social-evaluative anxiety. In T. V. Merluzzi, C. R. Glass, & M. Genest (Eds.), Cognitive assessment. New York: Guilford. Halford, W. K., & Sanders, M. R. (1988). Assessment of cognitive self-statements during marital problemsolving: A comparison of two methods. Cognitive Therapy and Research, 12(5), 515–530. Houston, B. K., Fox, J. E., & Forbes, L. (1984). Trait anxiety and children’s state anxiety, cognitive behaviors, and performance under stress. Cognitive Therapy and Research, 8(6), 31–41. Kendall, P. C., & Chansky, T. E. (1991). Considering cognition in anxiety-disordered children. Journal of Anxiety Disorders, 5, 167–185. Kendall, P. C., & Ingram, R. (1987). The future for cognitive assessment of anxiety: let’s get specific. In L. Michelson & L. M. Ascher (Eds.), Anxiety and stress disorders: Cognitive-Behavioral Assessment and Treatment (pp. 89–104). New York: Guilford. Kendall, P. C., & Ronan, K. R. (1990). Assessment of children’s anxieties, fears, and phobias: Cognitivebehavioral models and methods. In C. R. Reynolds & R. W. Kamphaus (Eds.), Handbook of psychological and educational assessment of children. New York: Guilford. Lockyer, J. (1990). Short maths 2. New Zealand: Longman Paul. Lodge, J., Harte, D. K., & Tripp, G. (1998). Children’s self-talk under conditions of mild anxiety. The Journal of Anxiety Disorders, 12, 153–176. Manly, B. F. J. (1992). The design and analysis of research studies. Cambridge: Cambridge University Press. Martzke, J. S., Andersen, B. L., & Cacioppo, J. T. (1987). Cognitive assessment of anxiety disorders. In L. Michelson & L. M. Ascher (Eds.), Anxiety and stress disorders: Cognitive-behavioral assessment and treatment. New York: Guilford. Meichenbaum, D., & Cameron, R. (1981). Issues in cognitive assessment: An overview. In T. V. Merluzzi, C. R. Glass, & M. Genest. Cognitive assessment. New York: Guilford. Merluzzi, T. V., & Boltwood, M. D. (1989). Cognitive assessment. In A. Freeman, K. M. Simon, L. E. Beutler, & H. Arkowitz (Eds.), Comprehensive handbook of cognitive therapy. New York: Plenum. Norusis, M. J., & SPSS Inc. (1994). SPSS advanced statistics !6.1. Chicago: SPSS Inc. Prins, P. J. M. (1985). Self-speech and self-regulation of high- and low-anxious children in the dental situation: An interview study. Behavior Research and Therapy, 23(6), 641–650. Prins, P. J. M. (1986). Children’s self-speech and self-regulation during a fear-provoking behavioral test. Behavior Research and Therapy, 24(2), 181–191. Prins, P. J. M., Groot, M. J. M., & Hanewald, G. J. F. P. (1994). Cognition in test-anxious children: The role of on-task and coping cognition reconsidered. Journal of Consulting and Clinical Psychology, 62(2), 404–409. Prins, P. J. M. & Hanewald, G. J. F. P. (1997). Self-statements of test-anxious children: Thought-listing and questionnaire approaches. Journal of Consulting and Clinical Psychology, 65(3), 440–447. Sanders, M. R., & Dadds, M. R. (1992). Children’s and parents’ cognitions about family interaction:

418

Lodge, Tripp, and Harte

An evaluation of video-mediated recall and thought-listing procedures in the assessment of conductdisordered children. Journal of Clinical Child Psychology, 21(4), 371–379. Schwartz, R. M., & Garamoni, G. L. (1986). A structural model of positive and negative states of mind: Asymmetry in the internal dialogue. Advances in cognitive-behavioral research and therapy, 5, 1–62. Spielberger, C. D. (1973). State-Trait Anxiety Inventory for Children. Palo Alto, CA: Consulting Psychological Press. Tripp, E. G., & Lodge, J. F. (1993). Culturally sensitive test use—Let’s begin at home. Poster presented at the International Conference on Test Use with Children and Youth, Oxford, England. Wechsler, D. (1981). The Wechsler Adult Intelligence Scale—Revised. San Antonio: The Psychological Corporation. Zar, J. H. (1984). Biostatistical analysis (2nd ed.). Engewood Cliffs, NJ: Prentice-Hall International, Inc. Zatz, S., & Chassin, L. (1983). Cognitions of test-anxious children. Journal of Consulting and Clinical Psychology, 51(4), 526–534. Zatz, S., & Chassin, L. (1985). Cognitions of test-anxious children under naturalistic test-taking conditions. Journal of Consulting and Clinical Psychology, 53(3), 393–401.