STATISTIK-FORUM.de

[vux]

Hallo,

ich muss eine ähnliche Analyse machen, wie die die unten Beschrieben wird (Auszug aus einem A-Journal Paper, Entrepreneurship).

Dabei würde ich gerne zunächst wissen, was es heißt die abhängige Variable in einem Sample zu "residualisieren und standardisieren" mit den wichtigen Kontrollvariablen (siehe 1. bold Text unten).

Bei der Regression um die Isoquanten zu erstellen (deren Steigung die Austauschbarkeit der zu untersuchenden Variablen angibt), verstehe ich nicht was mit dem Ausdruck "regress Var A on Var B" (2.bold Text unten). Heißt dass einfach A ist die independent und B die dependent var im Regressionsmodell - ist das eine 1:1 Regression, d.h. nur 1 Independent?

Hoffe dass meine Anfrage verständlich ist, bei Nachfragen gerne hier mailen.

Danke!!!

Substitutability of Founders’ Human and Financial Capital
Although some variance is explained by differences in industry financial requirements,
there is much within industry variability that remains unexplained. Our second hypothesis
tested the substitutability of human and financial capital. First, we residualized and
standardized the growth and sales/earnings measures by business age and business type
to control for age and industry differences. We then split the sample based on the medians
for human and financial capital (residualized by age and industry). This resulted
in four groups: (1) low human capital-low financial capital, (2) low human capital-high
financial capital, (3) high human capital-low financial capital, (4) high human capitalhigh
financial capital. We used ANOVA followed by Scheffe´ ’s test, to test for differences
in sales/earnings and growth between groups. No significant differences in growth
rates existed between groups. However, theANOVA with sales/earnings as the dependent
variable was significant (F56.62, p,.01). The groups with high levels of financial
and human capital had significantly (Scheffe´ ’s, p , .01) higher levels of sales/earnings
than all other groups. The mean sales/earnings of the high human capital-low financial
capital and low human capital-high financial capital groups were virtually identical. This
finding provides some support for our second hypothesis that human and financial capital
are at least partial substitutes.

We sought to develop greater insights into the degree of substitutability between
human and financial capital by developing isoquant curves. To do so we focused on the
two subgroups (high human capital, low financial capital, and low human capital, high
financial capital) where substitution appears to occur.We developed isoquants by looking
specifically at two sales/earnings levels: those in the top sales/earnings quartile and
those in the bottom sales/earnings quartile after residualizing for business age and business
type. We did not expect perfect functions for the following reasons: (1) even
though the companies have been placed in similar sales/earnings groups, there is still
some sales/earnings variance and therefore firms are not producing identical output
quantities, (2) our measures of human and financial capital are relatively crude, (3) we
can not hold everything constant as is suggested in economic models, and (4) although
we have controlled for business type in the sales/earnings, different firms within industries
do not use identical production functions.
Because of these restrictions we do not expect a perfect fit, yet significant results
will validate the application of the isoquant model. To develop these curves we regressed
human capital on financial capital for both of the above defined isoquant groups.We tested for curvilinearity, but found only linear solutions to be significant. At both
isoquants, the slope is negative and significant providing support for hypothesis 2. The
slopes of the isoquant curves provide insights into the substitutability of resources. The
slope of the isoquant curve generated for the top sales/earnings quartile is 254,333,
indicating that each unit of human capital (enough experience to move from one level
to the next in the scale described earlier, roughly 2–3 years of experience) is substitutable
for $54,333 of financial capital, with a 95% confidence interval ranging from $33,273
to $75,393. Likewise, the slope of the isoquant curve for the low sales/earnings group
is 231,003, indicating that each unit of human capital is substitutable for $31,003, with
a 95% confidence interval ranging from $24,427 to $37,579. of measurement, these slopes are not significantly different, suggesting the possibility
of parallel solutions as described in economic theory. However, their magnitude gives
some indication of the relative value of the stock of human capital as measured by experience
indicators. Results are displayed graphically in Figure 1.

[PonderStibbons]

Dabei würde ich gerne zunächst wissen, was es heißt die abhängige Variable in einem Sample zu "residualisieren und standardisieren" mit den wichtigen Kontrollvariablen (siehe 1. bold Text unten).

Normalerweise sollte das heißen, Regression mit Alter und Unternehmenstypen als Prädiktoren rechnen, Residuen der abhängigen Variable standardisieren, mit den standardisierten Residuen weitermachen.

"regress Var A on Var B" (2.bold Text unten). Heißt das einfach, A ist die independent und B die dependent var im Regressionsmodell - ist das eine 1:1 Regression, d.h. nur 1 Independent?

Man regrediert (führt zurück) eine abhängige auf eine unabhängige Variable, hier human capital (AV) on financial capital (UV). Was genau da abgeht, kannn ich nicht sagen, ist mir zu viel Text.

Gruß

P.