Enter your data for Augmented Dickey-Fuller Test

Stat > Time Series > Augmented Dickey-Fuller Test

In This Topic

Maximum lag order for terms in the regression model
Criterion for selecting lag order
Additional terms
Display plots for

In Series, enter a column of numeric data that were collected at regular intervals and recorded in time order.

This worksheet shows the first few rows of a table with 50 rows. Sales A contains the number of computers that are sold each month and is valid with Augmented Dickey-Fuller Test because the column has no missing values. Sales B is also valid with Augmented Dickey-Fuller Test because the column has only missing values at the beginning. Columns with missing values at the end or with missing values at the beginning and end are also valid. Sales C is not valid with Augmented Dickey-Fuller Test because a missing value is between two numbers in the series.

C1	C2	C3
Sales A	Sales B	Sales C
195000	*	195000
213330	195000	*
208005	213330	213330
249000	249000	208005
237040	237040	249000
…	…	…

Maximum lag order for terms in the regression model

Enter the highest lag order to evaluate when fitting the regression model to calculate the test statistic. By default, the final highest lag order is the order for the regression model that minimizes the Akaike Information Criterion (AIC).

Criterion for selecting lag order

Select a criterion to determine the final highest lag order for the regression model that the analysis uses to calculate the test statistic. Usually, the default criterion of Minimum AIC works well.

Minimum AIC: The selected lag order in the regression model minimizes the AIC.
Minimum BIC: The selected lag order in the regression model minimizes the Bayesian Information Criterion (BIC).
t-statistic: The selected lag order in the regression model is the highest lag order term with a significant t-statistic.
None: The analysis uses all lag orders less than or equal to the maximum lag order to fit the regression model.

Additional terms

Include additional terms in the regression model. The default is to include a constant term which says that the mean of the series is non-zero if the series is stationary.

Constant: Include a constant term in the model.
Constant + linear trend: Include a constant term and a linear trend in the model.
Constant + linear trend + quadratic trend: Include a constant term, a linear trend, and a quadratic trend in the model.
None: Include no additional terms in the model so that the mean of the series is 0 if the series is stationary.

Display plots for

Select whether to show the original time series data, the differenced time series data, or both series of data in plots. Select Time series, autocorrelation, and partial autocorrelation to show the plots for the chosen series. Without this selection, the results do not include plots.

Original and differenced series: Display plots for the original data and the differenced data.
Original series: Display plots only for the original data.
Differenced series: Display plots only for the differenced data.