Construct validity and between group differences

Construct validity is tested by evaluating hypotheses in relation to the measure. This might involve confirming or refuting hypotheses (Cronbach and Meehl 1955). Construct validity involves

collecting verifiable evidence to support the inference that a measure has meaning and is the most important approach to validity where there is no bench mark test (Wilkin et al 1992; Hays et al 1998; Kline 1999). The term ‘construct’ is similar to the term ‘concept’ (Kline 1999) and should be

explained. The construct or concept of dyspraxia/DCD relates to impairment in motor coordination in the absence of a congenital or acquired neurological disorder that occurs in childhood and which may continue into adulthood (See 4.1). Construct validity of the Functional Difficulties Questionnaire (FDQ-9) was explored in relation to this concept by employing the known groups method. It was expected that individuals who reported experiencing coordination difficulties in everyday life as an adult and as a child or who self-reported dyspraxia would score more highly on the FDQ-9 than those who reported no coordination difficulties or dyspraxia.

Participants

Construct validity was carried out employing data from S4 (n=152) because this group had been asked questions relevant to construct validity.

1. Methods

Participants from S4 (staff and students of a university) were asked an additional question - ‘Have you ever considered yourself to be ‘clumsy’ or uncoordinated in your everyday life?’ Responses were divided into the constructs; ‘yes difficulties as a child and as an adult’; ‘yes difficulties as a child only’; ‘yes difficulties as an adult only’ or ‘no difficulties’. Box-and-whisker plots are presented for the four constructs versus the Functional Difficulties Scores (FDS). This data is presented in figure 4-6.

Figure 4-6 Box-and-whisker plots for participants who reported functional difficulties in four sub groups of Sample 4 (n=152) and their FDS score

For each group the median, range of functional difficulty scores, Lower Quartile (LQ), Upper Quartile (UQ) and Inter Quartile Range (IQR) were reported.

 Those who reported ‘no’ difficulties (n=92) the median scores were 17, with a range of 11 – 25, LQ 14, UQ 19 and IQR 5. This would indicate that 75% of participants reported scores of 11-19. A score of 11 would broadly indicate reporting being ‘very good’ at 7/9 items and ‘good’ at 2/9 items. A score of 19 would broadly indicate reporting being ‘good’ at 8/9 items and ‘poor’ at 1/9 items.

 Those who reported ‘difficulties as a child and as an adult’ (n=36) the median scores were 22 with a range of 13 - 30, LQ 18, UQ 26 and IQR 8. This would indicate that 75% would report scores of 18 -30. A score of 18 would broadly indicate being ‘good’ at all 9 items, while as score of 30 would broadly indicate being ‘poor’ at 6/9 items and ‘very poor’ at 3/9 items.

 Those who reported difficulties ‘as a child only’ (n=12) the median scores were 18 with a range of 13 – 23, LQ 17, UQ 21 and IQR 4. This group align to the theoretical construct that not all children who report coordination difficulties in childhood continue to have difficulties in adulthood (Losse et al 1991; Cantell et al 1994).

 Those who reported ‘difficulties as an adult only’ (n=12) the median scores were 19 with a range of 14 – 23, LQ 14, UQ 22 and IQR 5. It is suggested that participants in this group who appear to acquire functional difficulties in adulthood do not have dyspraxia/DCD. It may be important to investigate this group further but is beyond the remit of this chapter in which the focus is on those who report difficulties in childhood and adulthood in line with the concept of dyspraxia/DCD in this study.

1. Methods

In the following analysis the known groups method was employed to investigate a hypothesis. It was anticipated that those who reported ‘difficulties both as a child and as an adult’ (n=36) would have a higher FDS than those who reported having ‘no difficulties’ (n=92). An independent samples unequal variance t-test was used to compare the mean FDQ-9 scores for these two groups. A Mann-Whitney U test was used to confirm this result.

1. Results

The mean score of those reporting ‘no difficulties’ (16.77 SD3.099) was significantly lower than those who reported ‘yes difficulties both as a child and as an adult’ (22.22 SD 4.517). Levene’s test,

p < 0.05 therefore unequal variances were assumed, t (48.45) = 6.653, p <0.001 (two tailed). This

result was further confirmed by employing the Mann-Whitney U test. Participants who reported ‘yes difficulties both as a child and as an adult’ reported significantly higher total scores than participants who reported ‘no difficulties’ U = 531.000, p < 0.001 (two tailed). The mean differences between the groups were -5.450 [95% CI – 3.804 to – 7.097]. This is a statistically significant difference and the

hypothesis was upheld.

In addition a difference of five points is likely to be clinically important. On average participants who reported ‘difficulties as a child and as an adult’ broadly recorded scores relating to being ‘poor’ in 4/9 items, indicative of reporting functional difficulties in four items . On average participants who reported ‘no difficulties’ broadly recorded scores relating to being ‘good’ in 8/9 items and ‘very good’ in 1/9 items, indicative or reporting no functional difficulties.

Participants from S4 were given the opportunity to self-report dyspraxia. The term dyspraxia was employed instead of DCD as the term DCD has only more recently been recognised (Polatajko et al 1995) and a diagnosis is usually confirmed in the first decade of life which would have been before the term DCD was in common use. A diagnosis of dyspraxia/DCD has only more recently been available for adults by employing the ADC (Kirby et al 2010) but this questionnaire was not published until after the data collection in this study. The data relating to self-report of dyspraxia versus FDS is presented in figure 4-7.

Figure 4-7 Box-and-whisker plots for participants who either self reported dyspraxia or did not by showing their Functional Difficulties Score (FDS) in Sample 4 (n=152)

For both groups the median, range of FDS, Lower Quartile (LQ), Upper Quartile (UQ) and Inter Quartile Range (IQR) are reported.

 Those who self-reported dyspraxia (n=7) the median FDS were 28, with a range of 20 – 30, LQ 22, UQ 29 and IQR 7.

 The range of FDS were smaller than that noted in those who did not self-report dyspraxia and indicate that all participants who self-reported dyspraxia broadly reported being ‘poor’ in at least 2/9 items.

 In addition 75% of participants report being ‘poor’ at 4/9 or more items. None of the participants reported being ‘good’ or ‘very good’ at all of the 9 items.

 Those who did not self-report dyspraxia (n=145) the median FDS were 18 with a range of 11 - 30, LQ 15, UQ 21 and IQR 6.

 These results indicate 25% of participants reported being ‘good’ or ‘very good at all 9 items.  In addition 25% of participants had FDS of between 21 and 30. Scores of 21 or more

indicate being ‘poor’ in at least 3/9 items. FDS of 30 indicate being ‘poor’ in 6/9 and ‘very poor’ at 3/9 items.

 Those with high FDS may be indicative of participants who were not assessed in their early years for dyspraxia/DCD. This is a group that has been previously identified and discussed (Kirby et al 2008).

2. Methods

In the following analysis the known groups method was employed to investigate a hypothesis. It was anticipated that those who self-reported dyspraxia (n=7) would have higher FDS than those who did not self-report dyspraxia (n=145). An independent samples equal variance t-test was used to compare the mean FDQ-9 scores for these two groups. A Mann-Whitney U test was used to confirm this result.

2. Results

The mean scores of those self reporting dyspraxia (25.86, SD 4.100) were significantly higher than those who did not self-report dyspraxia (18.06, SD3.777), Levene’s test, p > 0.05, therefore equal variances were assumed t(150) = 5.314 , p = 0.02 (two tailed). This result was further confirmed by employing the Mann-Whitney U test. Participants who reported dyspraxia reported significantly higher total scores than participants who did not report dyspraxia, U = 83.000, p < 0.001 (two tailed). The mean differences between the groups were 7.795 [95% CI 3.998 to – 11.593]. These results indicated a statistically significant difference and the hypothesis were upheld. In addition a mean difference in the scores of 8 points is likely to be clinically important. On average those participants who self-reported dyspraxia broadly recorded scores relating to being ‘poor’ in 7/9 items indicative of reporting functional difficulties in the majority of items. On average those who did not self-report dyspraxia broadly recorded scores relating to being ‘good’ at 9/9 items and therefore did not report any functional difficulties.