Lecture 6:

Moderation: the relationship between x and y depends on a third variable z, z is than the moderation. In a regression multiplicative interaction term between the predictor x and the moderator z. For making hypothesis: effects become stronger, weaker or change direction.

Moderation: statistical approach:

Regress depend variable on:

1. Predictor (x)
2. Moderator (z)
3. Interaction of the two (xz)

Tips:

Include interaction terms if hypothesis is conditional:

Conditional hypothesis:

• A relationship between two or more variables depends on the value of one or more other variables
• An increase in x is associated with an increase in y when condition z is met, but not when condition z is absent
• A positive effect of x on y gets stronger as z increases
• Conditional hypotheses can be tested using interaction terms

Include all constitutive terms: include each of the elements that constitute interaction term:

• If you include xz, always include x and y
• If you include xzj, always include x, z, j, xz, zj and xj
• If you include X^2, always include x

Interpret correctly:

Interpretation of B is now different, so B2 does no longer explains the increase of x on y (Only when z is zero, it does) to see this you should calculate the marginal effect of x on y for different values of z

Meaningful MEs and SEs:

Correct interpretation of B’s involve derivation of marginal effects of the independent variables and the uncertainty with which they are estimated

Without interaction

With interaction:

Traditional results tables only report Bs and SEs, but we can only tell whether x has an significant effect on y when z is 0, however this could also be on other variables.

If z is binary present four number:

1. Marginal effect of x when z is 0 and z is 1
2. The corresponding standard errors

If z is continuous a graphical analysis is required:

1. Plot the marginal effect of x
2. Across a substantively meaningful range of z
3. And add the confidence interval for assessing significance
4. Then  report the % of sample that falls within the region of significance (show histogram)

What is mediation:

The impact of x on y is driven by the presence of another variable, example, w is mediator and then the explanatory mechanism that drives the relation between x and y.

Moderation: conditional

Mediation: the relationship runs via a third variable w, y depends on mediator and mediator depends on x. Can be causcal result on x. Mediator explains process

Article:

• Firms compete an a global basis, expand abroad
• But there are risks and costs
• Uncertainty when in an unfamiliar environment
• How the firms responds to this determines its success
• That’s why we need to understand internationalization
• To minimize uncertainty risks in internationalization (incremental approach, gradual expansion)
• First enter markets with similar cultures, then dissimilar
• First exports, then local sales subsidiaries, then production facilities à incremental model of international expansion, but with mixed empirical support
• There must be conditions that restricts the theory power, identify them to understand mixed results
• Focus: firm’s internationalization process.
• Objectives:

1. Test whether foreign market uncertainty leads to firms to adopt an incremental approach
2. Examine the impact of firm, industry and host country factors on the internationalization process

• Explain conflicting findings in literature
• Hypotheses:

1. Firms that face greater foreign market uncertainty (FMU) will be more involved in incremental international expansion (IIE). Because of lack of foreign markets knowledge (language, culture, differences) and foreign markets knowledge is tacit and can only be gained through experience. Is supported by results
2. Organizational slack (OS), Resource recoverability (RR) and locational advantage (LA), weaken the positive effect of H1. The higher the OS, the less positive the relationship between FMU and firm’s involvement in IIE. Slack excess resources shield firms from risks and allow firms to experiment with new strategies. The higher the RR, the less positive the relationship between FMU and firms involvement in IIE. High RR, lower sunk costs, resources can be used for other purposes, less risk. The higher the LA, the less positive the relationship between  FMU and firm’s involvement in IIE. Opportunities (costs of production). This is conditional hypothesis, the moderator is LA.

• Sample, data, methods:
• We need data on:

1. IIE
2. FMU (Cultural uncertainty and operation uncertainty
3. OS, RR, LA
4. Control variables: Firm size, firm age and market size

• Source: firms reports, interviews with managers
• Sample: manufacturing firms, south Korea, complete history of international expansion, at least one overseas manufacturing facility. Period between 1954-1996 and 19 host countries. For analysis use regression
• Hypothesis 2 is not supported, hypothesis 3 partiality supported. Hypothesis 3 also partially supported.
• For marginal error: if ^2 it is 2x b4