Productive play time: the effect of practice on consumer demand for hedonic experiences

Note that if it were difficult to convert a measure of K into the same units as N so they could be summed, potentially a different slope could be estimated for each kind of practice.

As we report below, the pattern of results was not substantially different whether we measured demand with three items (including intention) or with just the two evaluative items; however, conceptually (as explained above), intention is an important component of demand.

Our results were virtually identical when we deleted the intention item from our measure of demand and instead used a measure based on the two evaluative items (r = .89; Spearman Brown split half reliability = .94). The main effect of practice (β = .51, p = .001) was qualified by a marginally significant two-way interaction between practice and prior knowledge (β = −.30, p = .055). There was also a marginally significant interaction between prior knowledge and difficulty (β = −.30, p = .056), but the three-way interaction between practice, prior knowledge, and difficulty was not significant (β = .09, p = .557).

Following the suggestion of an anonymous reviewer, we confirmed this test of H3 using a two-group structural equation model (SEM), in which demand was a latent variable and the two groups were high and low prior knowledge. Using Anderson and Gerbing’s (1988) two-step approach, the first-step’s unconstrained model had satisfactory fit indices (χ²(18) = 4.61, p = .100, GFI = .99, NFI = .99, CFI = .99, RMSEA = .08), although ideally RMSEA should be below .05. In the second step, the fit of the model was not significantly worse after constraining the measurement weights to be equal in both groups (χ²(16) = 4.15, p = .272, GFI = .99, NFI = .99, CFI = 1.00, RMSEA = .04; difference χ²(2) = .54, p = .765). However, further constraining the structural path from practice to demand to be equal for both groups did significantly affect fit (difference vs. measurement weights model: χ²(1) = 7.11, p = .008). Practice had a significant positive effect only for the low prior knowledge group (standardized β = .51, p = .002, vs. high prior knowledge β = .10, p = .369).

The three-way interaction between practice, difficulty, and prior knowledge predicted by H4 was also significant when we used a two-item measure of demand that did not include the intention item (β = −.82, p = .017). The main effect of practice was negative (β = −.84, p = .025), as was the main effect of perceived difficulty (β = −.45, p = .009), but both these main effects were qualified by the significant three-way interaction, and also a significant two-way interaction between practice and difficulty (β = 1.04, p = .002).