Students' attitudes toward statistics: implications for the future

College Student Journal, Sept, 2004 by Jamie D. Mills

Logit[(P(Y < j)] = [alpha]j [beta]x,

where j = 1.... J-1 and the parameter [beta] describe the effect of x on the log odds of response in category j or below. The cumulative probabilities are the probabilities that the response Y falls in category j or below, for each possible j. For a 2 group scenario, the model fits well when subjects in one group tend to make higher responses on the ordinal scale than subjects in the other group (Agresti, 1996).

Local odds ratios for pairings of categories were used to describe the model, when appropriate. Local ratios use adjacent rows and columns. For the 2 X 2 table using cells intersecting rows a and c with columns b and d, the model has odds ratio equal to

[mu]ab [mu]cd / [mu]ad [mu]cb = exp[[beta]([u.sub.c] - [u.sub.a]) ([v.sub.d] - [v.sub.b]),

where the row and column numbers were the scores assigned.

Results

For this study, Cronbach coefficient alpha reliability estimates were generated by subscale. For the affect subscale consisting of 6 items, the coefficient alpha was .85 (n = 201). The coefficient reliability estimates for the cognitive competence (6 items), value (8 items), and difficulty (7 items) scales were .80 (n = 203), .88 (n = 202), and .71 (n = 202), respectively. The four subscales provided acceptable reliability estimates.

Tables 1-3 present the descriptive statistics by subscale for the SATS (please see the Appendix for a copy). In Table 1, items 1, 2, 10, 13, 14, and 20 measured affect while items 3, 8, 19, 22, 23, and 26 comprised the cognitive competence subscale. The value subscale was measured by items 5, 7, 9, 11, 12, 15, 18, and 24 and items 4, 6, 16, 17, 21, 25, and 27 measured difficulty.

For the affect subscale, the mode response of 4 indicated that student attitudes were neutral toward affect. An investigation of modal responses for items on this subscale revealed that students both agreed and disagreed with statements from this subscale. For example, students agreed to the statement 'I like statistics' (mode = 5) but they disagreed that they enjoyed taking statistics courses (mode = 3) and agreed that they get frustrated over statistics tests in class (mode = 5). Students also disagreed to the statement 'I feel insecure when I have to do statistics problems' (mode = 3) and they strongly disagreed that they were scared by statistics (mode = 1).

An overall sentiment of disagreement (mode = 2) described the student attitudes for the cognitive competence scale. Although students strongly agreed that they can learn statistics (mode = 7) and agreed that they can understand statistics equations (mode = 5), they disagreed that they have trouble understanding statistics because of the way they think (mode = 2), that they make a lot of math errors (mode = 2), and they strongly disagreed that they have no idea what's going on in statistics (mode = 1). They also disagreed that they find it difficult to understand statistics concepts (mode = 3).

The typical sentiment for the items on the value subscale was also disagreement (mode = 2). Students disagreed that statistics is worthless (mode = 1), statistical thinking is not applicable outside of their job (mode = 2), statistics is not useful to the typical profession (mode = 2), statistics is; not applicable in their profession (mode = 2), statistics conclusions are rarely presented in everyday life (mode = 3), and statistics is irrelevant in one's life (mode = 3). Students agreed that statistics should be required as a part of their professional training (mode = 5) and that they use statistics in their everyday lives (mode = 5).